Research On Road DamageIdentification Method Based On Feature Enhancement Of Laser Point Cloud

Intelligent acquisition of road damage information is a great significance to road planning,construction and maintenance,and have important means to improve the intelligent management of roads.The MMS can obtain the three-dimensional point cloud information of the road,which provides a new data support for the road damage extraction.In this paper,based on the data of vehicle laser point cloud,a method of road damage extraction based on feature enhancement is studied.The main problems are as follows:the three-dimensional point cloud data are distributed discretely,and the direct processing efficiency is low;the shape characteristics of the edge and the internal area of the pavement damage are different,which is difficult to express structurally.Based on the above problems,this paper carried out in-depth research and the intelligent recognition of road damage by enhancing the characteristics of laser point cloud.The main contents are as follows:1.Feature image generation.Based on the difference of the elevation gradient of the damaged edge and the randomness of the distribution direction of the damaged edge,a circular structure elevation gradient difference operator is constructed to generate the road gradient characteristic map;then,based on the difference of the spatial distance between the inner area of the damaged object and the surrounding road,the reference plane fitting algorithm of the local area of the road is studied to generate the road depth characteristic map.2.Pavement damage division.In view of the road gradient feature map,the filter method under the constraint of feature extremum is used to remove salt and pepper noise,and a dynamic window mode is designed to segment the damaged edge features.Then,based on the road depth feature map,according to the statistical spatial distance information,the normal distribution is used to calculate the threshold value dynamically to segment the damaged internal area features.3.Pavement damage fusion.Based on the two kinds of feature maps,the edge features and inner area features of road damage are extracted accurately.The two kinds of feature pixel values after binarization are superposed on the real coordinates,so that the damage information can be completely expressed.Then,morphological and binary clustering operators are used to remove the combined block noise and generate complete road damage vector elements.The final road damage target extracted by this method is transformed into an editable vector element result.Through two groups of different types of road data,the experimental demonstration and quantitative analysis of this algorithm are carried out.The average accuracy rate of damage extraction is 89.47%,and the average recall rate is 90.55%,which meets the actual production demand.It is verified that this algorithm has a good recognition of road damage in the road effect.